Re: [patch] Re: PostgreSQL pgbench performance regression in 2.6.23+

[Mike Galbraith <efault@gmx.de>]

On Fri, 2008-06-06 at 08:13 +0200, Mike Galbraith wrote:
> On Fri, 2008-06-06 at 01:03 -0400, Greg Smith wrote:
> 
> > I think I might not be testing exactly the same thing you did, though, 
> > because the pattern doesn't match.  I think that my Q6600 system runs a 
> > little bit faster than yours, which is the case for small numbers of 
> > clients here.  But once we get above 8 clients your setup is way faster, 
> > with the difference at 15 clients being the largest.  Were you perhaps 
> > using batch mode when you generated these results?
> 
> No, those were with stock settings.
> 
> > Regardless, clearly your patch reduces the regression with the default 
> > parameters to a mild one instead of the gigantic one we started with.
> 
> Unfortunately, after the recent reverts, we're right back to huge :-/
> 
> I'm trying to come up with a dirt simple solution that doesn't harm
> other load types.

The below doesn't hurt my volanomark numbers of the day, helps pgbench
considerably, and improves the higher client end of mysql+oltp a wee
bit.  It may hurt the low end a wee bit, but the low end is always
pretty unstable, so it's hard to tell with only three runs.

pgbench
2.6.26-rc5                         2.6.26-rc5+
1  10213.768037    10237.990274    10165.511814    10183.705908
2  15885.949053    15519.005195    14994.697875    15204.900479
3  15663.233356    16043.733087    16554.371722    17279.376443
4  14193.807355    15799.792612    18447.345925    18088.861169
5  17239.456219    17326.938538    20119.250823    18537.351094
6  15293.624093    14272.208159    21439.841579    22634.887824
8  12483.727461    13486.991527    25579.379337    25908.373483
10 11919.023584    12058.503518    23876.035623    22403.867804
15 10128.724654    11253.959398    23276.797649    23595.597093
20  9645.056147     9980.465235    23603.315133    23256.506240
30  9288.747962     8801.059613    23633.448266    23229.286697
40  8494.705123     8323.107702    22925.552706    23081.526954
50  8357.781935     8239.867147    19102.481374    19558.624434

volanomark
2.6.26-rc5
test-1.log:Average throughput = 101768 messages per second
test-2.log:Average throughput = 99124 messages per second
test-3.log:Average throughput = 99821 messages per second
test-1.log:Average throughput = 101362 messages per second
test-2.log:Average throughput = 98891 messages per second
test-3.log:Average throughput = 99164 messages per second

2.6.26-rc5+
test-1.log:Average throughput = 103275 messages per second
test-2.log:Average throughput = 100034 messages per second
test-3.log:Average throughput = 99434 messages per second
test-1.log:Average throughput = 100460 messages per second
test-2.log:Average throughput = 100188 messages per second
test-3.log:Average throughput = 99617 messages per second


Index: linux-2.6.26.git/kernel/sched_fair.c
===================================================================
--- linux-2.6.26.git.orig/kernel/sched_fair.c
+++ linux-2.6.26.git/kernel/sched_fair.c
@@ -664,6 +664,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
 
 	update_stats_dequeue(cfs_rq, se);
 	if (sleep) {
+		se->last_preempter = NULL;
 		update_avg_stats(cfs_rq, se);
 #ifdef CONFIG_SCHEDSTATS
 		if (entity_is_task(se)) {
@@ -692,8 +693,10 @@ check_preempt_tick(struct cfs_rq *cfs_rq
 
 	ideal_runtime = sched_slice(cfs_rq, curr);
 	delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime;
-	if (delta_exec > ideal_runtime)
+	if (delta_exec > ideal_runtime) {
+		curr->last_preempter = NULL;
 		resched_task(rq_of(cfs_rq)->curr);
+	}
 }
 
 static void
@@ -994,6 +997,7 @@ wake_affine(struct rq *rq, struct sched_
 	    unsigned int imbalance)
 {
 	struct task_struct *curr = this_rq->curr;
+	struct sched_entity *se = &curr->se, *pse = &p->se;
 	unsigned long tl = this_load;
 	unsigned long tl_per_task;
 	int balanced;
@@ -1002,14 +1006,26 @@ wake_affine(struct rq *rq, struct sched_
 		return 0;
 
 	/*
+	 * If the current task is being wakeup preempted by multiple tasks
+	 * that it awakened, such that it can't get significant work done
+	 * between preemptions, try to spread these preemption sources.
+	 */
+	if (sync && se->last_preempter && se->last_preempter != pse) {
+		u64 se_last_exec = se->sum_exec_runtime - se->prev_sum_exec_runtime;
+
+		if (se_last_exec < sysctl_sched_migration_cost)
+			return 0;
+	}
+
+	/*
 	 * If sync wakeup then subtract the (maximum possible)
 	 * effect of the currently running task from the load
 	 * of the current CPU:
 	 */
 	if (sync)
-		tl -= current->se.load.weight;
+		tl -= se->load.weight;
 
-	balanced = 100*(tl + p->se.load.weight) <= imbalance*load;
+	balanced = 100*(tl + pse->load.weight) <= imbalance*load;
 
 	/*
 	 * If the currently running task will sleep within
@@ -1017,8 +1033,8 @@ wake_affine(struct rq *rq, struct sched_
 	 * woken task:
 	 */
 	if (sync && balanced && curr->sched_class == &fair_sched_class) {
-		if (curr->se.avg_overlap < sysctl_sched_migration_cost &&
-				p->se.avg_overlap < sysctl_sched_migration_cost)
+		if (se->avg_overlap < sysctl_sched_migration_cost &&
+				pse->avg_overlap < sysctl_sched_migration_cost)
 			return 1;
 	}
 
@@ -1219,8 +1235,27 @@ static void check_preempt_wakeup(struct 
 		pse = parent_entity(pse);
 	}
 
-	if (wakeup_preempt_entity(se, pse) == 1)
-		resched_task(curr);
+	if (wakeup_preempt_entity(se, pse) == 1) {
+		int preempt = 1;
+
+		/*
+		 * If current task is being prempted by multiple wakees,
+		 * tag it for 1:N affine wakeup preemption avoidance.
+		 */
+		if (se->last_preempter && se->last_preempter != pse &&
+				se->load.weight >= pse->load.weight) {
+			u64 exec = se->sum_exec_runtime - se->prev_sum_exec_runtime;
+
+			if (exec < sysctl_sched_migration_cost)
+				preempt = 0;
+		}
+
+		if (se == &current->se)
+			se->last_preempter = pse;
+
+		if (preempt)
+			resched_task(curr);
+	}
 }
 
 static struct task_struct *pick_next_task_fair(struct rq *rq)
Index: linux-2.6.26.git/include/linux/sched.h
===================================================================
--- linux-2.6.26.git.orig/include/linux/sched.h
+++ linux-2.6.26.git/include/linux/sched.h
@@ -963,6 +963,7 @@ struct sched_entity {
 
 	u64			last_wakeup;
 	u64			avg_overlap;
+	struct sched_entity	*last_preempter;
 
 #ifdef CONFIG_SCHEDSTATS
 	u64			wait_start;
Index: linux-2.6.26.git/kernel/sched.c
===================================================================
--- linux-2.6.26.git.orig/kernel/sched.c
+++ linux-2.6.26.git/kernel/sched.c
@@ -2176,6 +2176,7 @@ static void __sched_fork(struct task_str
 	p->se.prev_sum_exec_runtime	= 0;
 	p->se.last_wakeup		= 0;
 	p->se.avg_overlap		= 0;
+	p->se.last_preempter		= NULL;
 
 #ifdef CONFIG_SCHEDSTATS
 	p->se.wait_start		= 0;